Identifying ylide ions and methyl migrations in the gas phase: the decarbonylation reactions of simple ionized N-heterocycles1

Identifying ylide ions and methyl migrations in the gas phase: the decarbonylation reactions of simple ionized N-heterocycles1

International Journal of Mass Spectrometry 195/196 (2000) 71– 83 Identifying ylide ions and methyl migrations in the gas phase: the decarbonylation r...

278KB Sizes 2 Downloads 34 Views

International Journal of Mass Spectrometry 195/196 (2000) 71– 83

Identifying ylide ions and methyl migrations in the gas phase: the decarbonylation reactions of simple ionized N-heterocycles David J. Lavoratoa, Lorne M. Fella, Graham A. McGibbonb,1, Semia Senb, Johan K. Terlouwa,*, Helmut Schwarzb a Department of Chemistry, McMaster University, 1280 Main Street West, Hamilton, Ontario L8S 4M1, Canada Institut fu¨r Organische Chemie der Technischen Universita¨t Berlin, Straße des 17. Juni 135, D-10623 Berlin, Germany

b

Received 3 May 1999; accepted 8 July 1999

Abstract The decarbonylation reactions of ionized 2-acetylpyridine, 2-acetylpyrazine, and 2-acetylthiazole have been investigated using the multiple collision technique of neutralization–reionization/collision-induced dissociation mass spectrometry and 䡠⫹ related techniques. The resultant heterocyclic C6H7N䡠⫹, C5H6N䡠⫹ ions were identified as 2-methylene-1,22 , and C4H5NS 䡠⫹ 䡠⫹ dihydropyridine, 6 , 2-methylene-1,2-dihydropyrazine, 9 , and 2-methylene-2,3-dihydrothiazole, 12 䡠⫹, respectively. This result refutes proposals in the older literature that the decarbonylation would involve a methyl transfer yielding the 2-methyl species 2-methylpyridine (2 䡠⫹), 2-methylpyrazine (8 䡠⫹), and 2-methylthiazole (11 䡠⫹). Literature proposals for a methyl transfer in the dissociation of ionized dimethyl-2,3-pyridinedicarboxylate and methyl-4-pyridinecarboxylate were also examined, but could not be substantiated either. However, the proposed gas phase synthesis of the N-methylpyridinium ylide, 1 䡠⫹, from ionized methyl-2-pyridinethiocarboxylate did provide evidence for a genuine methyl migration. To reinforce these conclusions, the dissociation characteristics of isomers structurally related to 6 䡠⫹ (3 䡠⫹– 5 䡠⫹ and 7 䡠⫹), to 9 䡠⫹ (10 䡠⫹) and to 12 䡠⫹ (13 䡠⫹– 15 䡠⫹) were also considered. Exploratory quantum chemical calculations (at the B3LYP/6-31G* level of theory) indicate that 6 䡠⫹ and 12 䡠⫹ are among the most stable isomers in the C6H7N 䡠⫹ and C4H5NS 䡠⫹ systems, lying 24 and 19 kcal/mol lower in energy than 2 䡠⫹ and 11 䡠⫹, respectively. Their neutral counterparts, however, are considerably less stable: 6 is calculated to be 27 kcal/mol higher in energy than 2. Nevertheless, from neutralization–reionization experiments it follows that the neutral counterparts of the ionized decarbonylation products 6, 9, and 12 are stable molecules on the microsecond time scale. Significant 1,3-hydrogen shift barriers hinder the interconversion of both the ions and the neutrals into their 2-methyl substituted counterparts, thus accounting for their stability in the dilute gas phase. (Int J Mass Spectrom 195/196 (2000) 71– 83) © 2000 Elsevier Science B.V. Keywords: Neutralization–reionization mass spectrometry; Density functional calculations; Decarbonylation

1. Introduction A profusion of mass spectrometric methods is available for the determination of structures and

* Corresponding author. E-mail: [email protected] 1 Present address: Boehringer Ingelheim (Canada) Ltd, Bio-Me´ga Research Division, 2100 Cunard St., Laval Quebec, H7S 2G5, Canada. Dedicated to the memory of Professor R.R. Squires. 1387-3806/00/$20.00 © 2000 Elsevier Science B.V. All rights reserved PII S 1 3 8 7 - 3 8 0 6 ( 9 9 ) 0 0 1 4 7 - 5

elucidation of reaction pathways for gaseous ions and molecules [1]. Several of these, including tandem sector and Fourier transform ion cyclotron resonance (FTICR) mass spectrometries, have been aimed at characterizing six-membered ring containing Nheterocyclic C6H7N 䡠⫹ isomers including the methyl picolines 2 䡠⫹– 4 䡠⫹ [2] and the N- and 2-methylene pyridinium ylides 5 䡠⫹ and 6 䡠⫹ [3– 6]. The recognition

72

D.J. Lavorato et al./International Journal of Mass Spectrometry 195/196 (2000) 71– 83

Diagram 1.

of six structural isomers was not precluded by the limited number of distinguishing features in collisioninduced dissociation (CID) [7], charge-stripping (CS), and neutralization–reionization (NR) [8] mass spectrometry experiments [4 – 6]. However, despite the application of this battery of techniques and even chemometric methods of analysis [4,5], further innovation is still required to demonstrate to what extent the various isomeric ions may be capable of interconversion and to confirm that they have indeed stable (nonisomerizing) neutral counterparts for the less common C6H7N molecules. In this context, it should be noted that the radical cation, C5H4NCH3䡠⫹, 1 䡠⫹, has appeared among several proposals accounting for the dissociation of 2-substituted pyridines by methyl group transfer [9,10]. Furthermore, loss of CO from ionized 2-acetylpyridine has been reported [11,12] but

without any structural characterization of the product ion. In light of the attention devoted to the chemistry of other 2-keto pyridines [12] and related 2-carboxylic acid [13] and ester derivatives [14], it seems prudent to probe the C6H7N 䡠⫹ (m/z 93) ions from 2-acetylpyridine. There are additional systems immediately pertinent to the study of methyl migrations and structures of N-heterocycles. For example, closely related to the picoline radical cations are ionized methylpyrazine, 8 䡠⫹, and two of its distonic isomers, the 2-methylene pyrazinium ion 9 䡠⫹, and the 5-methylpyrazinium ion 10 䡠⫹. Another consists of the set of isomeric C4H5NS 䡠⫹ species including ionized 2-methylthiazole, 11 䡠⫹, the methylene dihydrothiazoles, 12 䡠⫹ and 13 䡠⫹, and the N-methyl and S-methyl ylide ions, 14 䡠⫹ and 15 䡠⫹.

Diagram 2.

D.J. Lavorato et al./International Journal of Mass Spectrometry 195/196 (2000) 71– 83

As a means to obtain additional structural and mechanistic insight on these N-heterocycles, we have employed the multiple collision technique of reionization/collision-induced dissociation (NR/CID) mass spectrometry [15]. The application of NR/CID as a structural assignment tool offers advantages in systems where isomeric ions (and neutrals) are not easily distinguished by the separate application of either CID and/or NR mass spectrometry. This may occur for several reasons: (1) mixtures of ions may be generated from ionized precursor molecules, (2) the isomeric ions (or neutrals) may interconvert either spontaneously or as a result of the collision process prior to fragmenting, (3) the ions may access only a limited number of dissociation channels, which do not result in fragment ions bearing unique mass-to-charge values, or (4) in NR experiments reionized neutral fragments may interfere. The third scenario may often be interrogated by performing tandem mass spectrometry (MS/MS) experiments, and through isotopic labeling experiments, whereas the occurrence of the first two is usually more difficult to identify. With this in mind, we report here an analysis of CID and NR/CID experiments on the above mentioned systems aided by selected ab initio calculations. Literature proposals involving methyl migration in substituted pyridine, pyrazine [16], and thiazole [17] radical cations have also been re-examined. On one hand, we were not able to find support for an earlier postulate for methyl transfer in the dissociation of dimethyl-2,3-pyridinedicarboxylate but on the other hand found evidence of the missing N-methylpyridinium ylide ion 1 䡠⫹ [10]. The decarbonylation of ionized 2-acetylpyrazine and 2-acetylthiazole appears to proceed analogously to that of the 2-acetylpyridine ion, yielding the 2-methylene radical cation derivatives rather than the 1-methyl or 2-methyl species. 2. Experimental procedure The tandem mass spectrometric experiments used to generate and characterize the C6H7N, C5H6N2, and C4H5NS ions and neutrals were performed at McMaster University on the VG Analytical ZAB-R. This BE1E2 (B ⫽ magnet, E ⫽ electric sector) instrument

73

and the details of the NR mass spectrum acquisition have been previously described [18], hence only a brief overview is provided here. The ions of interest, e.g. m/z 93 in the case of C6H7N 䡠⫹, are mass-selected by B and subsequently (in a small gas cell located between B and E1) subjected to collision with N,Ndimethylaniline. This results in neutralization by fast electron transfer of a fraction of the 10 keV source generated ions. After exiting the cell, the remaining ions are deflected by a positively charged electrode so that only a beam of fast moving neutral species enters a second gas cell. Here reionization takes place by collisions with oxygen molecules. The resulting ions are analyzed by scanning E1 to produce the conventional NR mass spectrum. If instead the reionized species are selectively transmitted at a fixed E1 setting, and oxygen gas is introduced in a third cell located between E1 and E2, a NR/CID mass spectrum characteristic of the reionized neutrals can be obtained by scanning E2. Comparative CID experiments are performed analogously but with the deflector electrode switched off. Spectra were recorded with a PC-based data system (Mommers Technologies Inc., Ottawa, Canada). The precursor molecules methyl-2-pyridinethiocarboxylate [10], bis(ethoxycarbonyl)-pyridinium methylide [19], 2-propylpyrazine [20], and methyl-5methylpyrazine-2-carboxylate [21] were synthesized according to literature procedures. All other compounds were purchased from commercial sources and used without further purification. Samples were introduced into the mass spectrometer via an all-glass heated inlet system equipped with a leak valve or a direct insertion-type probe having a glass bore and reservoir. At indicated pressures (monitored by a remote ionization gauge) of typically 10⫺6 Torr, ions were formed by electron ionization (70 eV) at a source temperature of 120 °C. The proton-bound dimer of 2-methylpyridine was produced in a chemical ionization source at a measured pressure of ⬃6 ⫻ 10⫺5 Torr. The computations were performed with the GAUSSIAN 94 series of programs [22] on workstations at the TU-Berlin and McMaster University. Geometries were optimized using the standard basis

74

D.J. Lavorato et al./International Journal of Mass Spectrometry 195/196 (2000) 71– 83

set 6-31G* and vibrational frequencies estimated with the popular hybrid density functional method B3LYP [23], employing Becke’s empirical three-parameter fit for mixing the Hartree-Fock and density functional exchange terms. All structures had no negative eigenvalues, and the data presented are corrected for zero-point vibrational energy (ZPVE) contributions. The ZPVE were scaled by a factor of 0.9806 [24] for the B3LYP fundamentals. The 具S 2 典 expectation value of the noninteracting Kohn-Sham determinant was in all cases very close to the exact values.

3. Results and discussion 3.1. Generation and characterization of C6H7N isomers As first reported by Ferretti and Flanagan [11a], dissociative electron ionization (EI) of 2-acetylpyridine produces in its normal EI mass spectrum ions at m/z 93 by loss of CO. To investigate the structure of the resulting C6H7N 䡠⫹ ions, i.e. to differentiate these ions from their isomers, the technique of collisioninduced dissociation was employed. Seven cyclic C6H7N 䡠⫹ isomers were generated via EI of the appropriate substituted pyridine precursors. The cations 2 䡠⫹– 4 䡠⫹ were realized by simple electron ionization of their neutral counterparts while 1 䡠⫹, 5 䡠⫹, 6 䡠⫹ and 7 䡠⫹ were obtained by dissociative EI of methyl-2pyridinethiocarboxylate [10], bis(ethoxy-carbonyl) pyridinium methylide [5,25], 2-propylpyridine [5], and 4-pyridylacetic acid, respectively. The CID mass spectra of the putative ions 1 䡠⫹– 7 䡠⫹ obtained in the third field free region of the three sector magnetic deflection instrument, are presented in Fig. 1. In basic agreement with the study of Flammang and co-workers [5], the spectra of the 2-methylpyridine ion 2 䡠⫹ [Fig. 1(a)] and its methylene isomer 6 䡠⫹ [Fig. 1(b)] are similar but they do exhibit characteristic differences [26]. Perhaps the most structure diagnostic of these is the CH2 loss peak at m/z 79 in the spectrum of the 2-methylene isomer 6 䡠⫹. Furthermore, several fragment ion clusters show differences in their relative abundance: for example the

m/z 26 (C2H2䡠⫹) : 28 (CH2N⫹) peak intensity ratio is 1.5 for 2 䡠⫹ and 0.8 for 6 䡠⫹. The enhanced m/z 28 (H–C§N–H⫹) intensity is characteristic of a pyridine ring bearing a N–H moiety [15,27]. The more intense charge stripping peak (⫹⫹) at m/z 46.5 in Fig. 1(b) is further revealing of the ylide structure of 6 䡠⫹. The N-methylene ylide ion 5 䡠⫹ [Fig. 1(e)] is characterized by an even lower m/z 78 (– CH3) : 79 (– CH2) peak intensity ratio, a result of the N–C bond proffered for cleavage. The arguments presented above also account for its fairly high m/z 26 : 28 peak intensity ratio, 1.75, and the intense charge stripping peak at m/z 46.5. The remaining isomers of conventional structure, 3- and 4-methylpyridine (3 䡠⫹ and 4 䡠⫹) exhibit, see Figs. 1(g) and 1(h), only weak doubly charged ion peaks and they are easily distinguished from 2 䡠⫹ on the basis of the reduced m/z 78 intensity [2], as is also observed in their normal mass spectra. However, differentiation of 3 䡠⫹ and 4 䡠⫹ on the basis of their CID mass spectra is clearly not possible. The decarbonylation reaction of the 2-acetylpyridine ion has been proposed to yield ion 1 䡠⫹, via a 1,3 methyl shift to the ring nitrogen [12]. However, the source generated product ions yield a CID mass spectrum which is superimposable upon that of Fig. 1(b), indicating that the ion generated is 6 䡠⫹. Next, the structure of the m/z 93 ions generated from metastable 2-acetylpyridine ions was probed. The metastable ion/collision-induced dissociation sequence provides a MI/CID mass spectrum, Fig. 1(c), identical to the CID mass spectrum of the source generated ions, Fig. 1(b), thus indicating that the decarbonylation yields isomerically pure ions 6 䡠⫹. A possible mechanism for the generation of 6 䡠⫹ from ionized 2-acetylpyridine (I 䡠⫹) is presented in Scheme 1 along with that proposed earlier for its generation from ionized 2-propylpyridine (II 䡠⫹) [5,12]. In both cases the initial step involves the transfer of a hydrogen radical to the ring nitrogen. For 2-propylpyridine this is followed by C2H4 loss via a McLafferty rearrangement whereas 2-acetylpyridine decarbonylates by an ipso [28] type rearrangement. We further note that the decarbonylation reaction is associated with a large reverse activation energy. Metastable 2-acetylpyridine ions lose CO with a

D.J. Lavorato et al./International Journal of Mass Spectrometry 195/196 (2000) 71– 83

75

Fig. 1. CID mass spectra (3ffr, 7 keV ions, collision gas O2) of C6H7N 䡠⫹ (m/z 93) ions generated from (a) ionized 2-methylpyridine, (b) ionized 2-acetylpyridine or 2-propylpyridine, (c) metastable 2-acetylpyridine ions, (d) metastable methyl-2-pyridinethiocarboxylate ions, (e) ionized bis(ethoxycarbonyl)pyridinium methylide, (f) metastable 4-pyridylacetic acid ions, (g) ionized 3-methylpyridine, and (h) ionized 4-methylpyridine.

kinetic energy release of 890 meV but this process is in competition with the loss of CH2¢C¢O (metastable peak intensity ratio 10:1). The loss of ketene yields the ␣-ylide ion of pyridine and the activation energy for this process is ⬃26 kcal/mol [15]. A similar value must obtain for the decarbonylation but its calculated

minimum energy requirement is 44 kcal/mol lower [29,30] and thus this process is indeed characterized by a high reverse activation energy. Evidence for a true methyl migration was eventually obtained, thereby validating the (unverified) proposal by Budzikiewicz and co-workers [10] for the

76

D.J. Lavorato et al./International Journal of Mass Spectrometry 195/196 (2000) 71– 83

Scheme 1.

formation of the N-methyl ylide ion 1 䡠⫹ from ionized methyl-2-pyridinethiocarboxylate, III 䡠⫹, as depicted in Scheme 2. Ions 1 䡠⫹ are abundantly generated from the low energy (metastable) ions III 䡠⫹ by loss of COS. The CID mass spectrum [Fig. 1(d)] of these ions is characterized by an intense CH3䡠 loss (m/z 78) and tell-tale peaks at m/z 42 (CH3NCH⫹) and m/z 15 (CH⫹ 3 ). In the methyl-d3 isotopomer m/z 42 is shifted to m/z 45 and m/z 15 to m/z 18, confirming the transfer of the intact methyl group. Budzikiewicz et al. [10] also proposed a methyl migration in the formation of the m/z 93 C6H7N 䡠⫹ ions from dimethyl-2,3-pyridinedicarboxylate ions IV 䡠⫹, as depicted in Scheme 2. However, loss of CO2

from m/z 137 does not figure in either the MI or the CID mass spectrum of the m/z 137 ions in Scheme 2. Moreover, high resolution measurements showed that the source generated m/z 93 ions largely consist of C5H3NO 䡠⫹ ions with only a small contribution of C6H7N 䡠⫹. Corroborating this, the mass spectrum of the dimethyl-d6 isotopomer displays peaks at m/z 93 (C5H3NO 䡠⫹) and m/z 97 (C6H3D4N 䡠⫹), with a ratio of 4:1. The CID mass spectrum of the labeled m/z 97 ions was obtained and despite the lack of reference spectra, it does indicate that a clean generation of the product ion shown in Scheme 2 is highly improbable. The remaining isomer studied, 7 䡠⫹, was generated from ionized 4-pyridylacetic acid, V 䡠⫹, as depicted in

Scheme 2.

D.J. Lavorato et al./International Journal of Mass Spectrometry 195/196 (2000) 71– 83

77

Scheme 3.

Scheme 3. Careful analysis of spectra obtained at various ion source temperatures indicated that the source generated ions of putative structure 7 䡠⫹ were contaminated with 4-methylpyridine ions 4 䡠⫹ originating from thermal decarboxylation of the parent acid. An unadulterated ensemble of ions 7 䡠⫹ could be obtained from the unimolecular dissociation of metastable 4-pyridylacetic acid molecular ions. Their CID mass spectrum, see Fig. 1(f), is characterized by the lowest m/z 26 : 28 peak intensity ratio of the set, 0.63, and a relatively intense peak at m/z 79 (– CH2) adjacent to a weak m/z 78 (– CH3) signal. The presence of charge stripping peaks at m/z 46.5 and 45.5 of comparable abundance completes its structure characterization.* To document another proposed methyl migration, the C6H7N 䡠⫹ ions produced from methyl-4-pyridinecarboxylate (VI 䡠⫹) were examined. In 1972, Neeter and Nibbering [32] postulated that the decarboxylation of VI 䡠⫹ in the metastable time frame yields the 4-methylpyridine ion, see Scheme 3. However, we surmise that this is not the case since the resulting MI/CID mass spectrum of the product ion (not shown) * Ions 7 䡠⫹ depicted in Diagram 1 do not contain a pyridine ring bearing a N–H moiety. Nevertheless, the m/z 28 (H–C§N–H⫹) ion intenstiy is enhanced in the CID spectrum and this has led a reviewer to propose that ions 7 䡠⫹ are ionized 4-methylene-1,4dihydropyridine. This possibility cannot be excluded, but we note that our proposal is based upon the structure diagnostic peak at m/z 54, rationalized by Flammang et al. [5] to originate from the direct 䡠 bond cleavage 7 䡠⫹ 3 H–C§N–CH¢CH⫹ 2 ⫹ H–C§C–CH2. From a MS/MS spectrum and comparison with reference spectra [31] we conclude the m/z 54 ions likely have the proposed structure. Moreover, their major CID dissociation involves loss of C2H2 and formation of m/z 28 (H–C§N–H⫹) and this may explain the enhanced m/z 28 intensity in the spectrum of 7 䡠⫹.

is virtually identical with that presented in Fig. 1(f), indicating that 7 䡠⫹ is the isomer produced. High resolution experiments showed that the source generated m/z 93 ions consist of a 1:1 mixture of C6H7N 䡠⫹ and C5H3NO 䡠⫹ ions. Thus, the CID mass spectrum (not shown) of these ions is that of a mixture and the structure of the weak source generated C6H7N 䡠⫹ ions remains unresolved. To further characterize the ions and to probe the stability of the elusive neutrals 5 and 6, Flammang and co-workers have also reported the NR spectra of 2 䡠⫹– 7 䡠⫹. One proposal from their study is that the intense “survivor ion” peaks in the spectra of 5 䡠⫹ and 6 䡠⫹ reflect the stability of their neutral counterparts. Evidence for this proposal comes from the presence of structure diagnostic signals in the NR spectra. However, consistent with our own observations (not shown), the intensity distribution of the various clusters of fragment ions in the NR spectra is considerably different from those in the CID spectra. One reason for this discrepancy (see Sec. 1) could be interference from reionized neutral fragments. To verify that these differences are not due to an isomerically impure initial beam of source generated ions and/or a partial isomerization upon neutralization, we have obtained the NR/CID spectrum of 5 䡠⫹ and 6 䡠⫹ along with that of the conventional isomer 2 䡠⫹. These spectra are presented in Fig. 2 along with the CID mass spectra (3ffr) of the corresponding source generated ions. For ions 5 䡠⫹ and 6 䡠⫹, the NR/CID and CID mass spectra [Fig. 2(a)–(d)] are closely similar, indicating (1) that the ions and their neutral counterparts are stable noninterconverting species in the gas phase, (2) that the initial ion beams consisted of isomerically

78

D.J. Lavorato et al./International Journal of Mass Spectrometry 195/196 (2000) 71– 83

Fig. 2. NR/CID and comparative CID mass spectra (3ffr, 10 keV ions, collision gas O2) of C6H7N 䡠⫹ (m/z 93) ions. (a), (b) NR/CID and CID spectra of C6H7N 䡠⫹ from ionized 2-acetylpyridine; (c), (d) NR/CID and CID spectra of C6H7N 䡠⫹ from ionized bis(ethoxycarbonyl)pyridinium methylide; (e), (f) NR/CID and CID spectra of ionized 2-methylpyridine.

pure ions, and (3) that the structure diagnostic portions of the NR spectra are obscured by contributions from reionized neutral fragments. Note that when comparisons are made between the NR/CID and CID mass spectra, the signal at m/z 66 (– H,C,N), which is largely of metastable origin and thus sensitive to the method of ion preparation, is not considered. Interestingly enough, it is the NR/CID [Fig. 2(e)] and the CID [Fig. 2(f)] mass spectra of ionized

2-methylpyridine that exhibit a difference in the m/z 26:28 peak intensity ratio, such that the NR/CID spectrum of 2 䡠⫹ could be argued to be that of a mixture of ions 2 䡠⫹ and 6 䡠⫹. Considering that the neutral counterpart of 2 䡠⫹ is much more stable than that of 6 䡠⫹ (see the following), the possibility can be excluded that part of the neutralized ions 2 isomerize into 6. Similarly, the possibility can be discarded that a significant fraction of the ions 2 䡠⫹ generated by

D.J. Lavorato et al./International Journal of Mass Spectrometry 195/196 (2000) 71– 83

electron impact isomerizes into the more stable species 6 䡠⫹, via a 1,3-H transfer. This is because the neutralization–reionization efficiency of 2 䡠⫹ is four times greater than that of 6 䡠⫹. It is therefore more likely that we are dealing with a significant degree of postcollisional isomerization upon collisional ionization of the neutral species 2. To test this hypothesis, a fast moving beam of neutral species 2 was produced in the second field free region from the 10 keV mass selected protonbound dimer of 2 via the collision-induced dissociation into [2 ⫹ H]⫹ (5 keV) ⫹ 2 (5 keV). By using the deflector electrode, all ionic species were removed from the beam, thus allowing only the fast moving neutrals 2 to collide with O2 in a second collision chamber, held at ⫹2 kV. The resulting 7 keV m/z 93 ions were selectively transmitted into the third field free region and subjected to CID. The resulting partial CID mass spectrum, Fig. 3(a), resembles the NR/CID spectrum presented in Fig. 3(b), indicating that the common collisional ionization event is largely responsible for the significant degree of postcollisional isomerization of 2 䡠⫹ into 6 䡠⫹. For comparison, the 7 keV CID mass spectrum of the source generated 2 䡠⫹ ions is presented in Fig. 3(c). The 1,3-H transfer involved in the isomerization 2 䡠⫹ 3 6 䡠⫹ undoubtedly imposes a high barrier (⬃50 kcal/mol) upon the reaction. However, one should also consider that ions 2 䡠⫹ may contain up to 75 kcal/mol internal energy before any dissociation process becomes possible [33,34]. An increase in the average internal energy resulting from a collisional ionization event [35] could promote significantly isomerization of 2 䡠⫹, thus accounting for the observed differences in the NR/ CID and CID mass spectra. Finally, since ion 1 䡠⫹ could only be obtained through the dissociation of metastable precursor ions, NR experiments were not feasible and thus its neutral counterpart remains unobserved. To complement the experimental observations ab initio calculations were performed on four selected isomers viz. 1 䡠⫹, 2 䡠⫹, 5 䡠⫹, and 6 䡠⫹ as well as their neutral counterparts. The results are summarized in Table 1. Not surprisingly, the most stable species among the ionic isomers is not ionized 2-methylpyri-

79

Fig. 3. Partial CID mass spectra (3ffr, 7 keV ions, collision gas O2) of the m/z 93 ions derived from (a) collisional ionization of 2-methylpyridine produced by the decomposition of its protonbound dimer (see text for details), (b) neutralization-reionization of 2-methylpyridine ions, and (c) 2-methylpyridine subjected to electron ionization.

dine 2 䡠⫹, but rather 6 䡠⫹, which lies 24 kcal/mol lower in energy. The N-methylene isomer 5 䡠⫹ is also more stable than 2 䡠⫹, by 11 kcal/mol, but 1 䡠⫹ is higher in energy, by 9 kcal/mol. These results are in fair agreement with a previous AM1 study [5] where 6 䡠⫹ and 5 䡠⫹ were found to be more stable than 2 䡠⫹, by 33 and 10 kcal/mol, respectively. For the neutral counterparts, the relative stability of 6 and 2 is reversed, with 6 predicted to be 27 kcal/mol higher in energy than 2, the most stable neutral of the set. Neutrals 1 and 5 are even higher in energy but, as could be experimentally verified for 5, they do represent stable species on the potential energy

80

D.J. Lavorato et al./International Journal of Mass Spectrometry 195/196 (2000) 71– 83

Table 1 Calculated energies for selected C6H7N and C4H5NS ions and neutrals at the B3LYP/6-31G* level of theorya Species

B3LYP

具S 2 典

ZPVEc

E relb

C 6 H 7 N 䡠⫹ isomers 1 䡠⫹ 2 䡠⫹ 5 䡠⫹ 6 䡠⫹

⫺287.278 36 ⫺287.289 11 ⫺287.309 82 ⫺287.330 99

0.778 0.756 0.775 0.770

75.1 73.3 74.6 75.3

9 0 ⫺11 ⫺24

C 6 H 7 N isomers 1 2 5 6

⫺287.519 85 ⫺287.605 79 ⫺287.531 37 ⫺287.562 54

74.4 74.6 73.9 74.5

54 0 46 27

C 4 H 5 NS 䡠⫹ isomers 11 䡠⫹ 12 䡠⫹ 13 䡠⫹d 14 䡠⫹ 15 䡠⫹

⫺608.054 34 ⫺608.086 32 ⫺608.029 55 ⫺608.011 00 ⫺607.961 66

50.0 51.5 50.6 51.5 49.7

19 0 35 42 77

0.759 0.769 0.776 0.761 0.759

a

Total energies in Hartree, relative and zero-point vibrational energies are in kcal/mol. Relative energies include the ZPVE. c ZPVE determined at B3LYP and scaled by 0.9806. d Note that the optimized geometry of this ion represents an open chain structure, see text. b

surface. Although the isomeric ions and neutrals have considerably different stabilities, they must be separated by sizable barriers which hinder a facile intramolecular isomerization. 3.2. Generation and characterization of C5H6N2 and C4H5NS isomers The decarbonylation of 2-acetylpyrazine and 2-acetylthiazole ions may be expected to occur analogously to that established for ionized 2-acetylpyridine. If so, the prominent m/z 94 and m/z 99 ions in their mass spectra should have the structures 9 䡠⫹ and 12 䡠⫹ rather than 8 䡠⫹ (2-methylpyrazine) and 11 䡠⫹ (2-methylthiazole), as proposed in the literature [16,17]. For the pyrazine system, this is confirmed by an analysis of the CID and NR/CID spectra presented in Fig. 4. As with the C6H7N 䡠⫹ system, the CID spectra of the ions generated by the loss of CO from the 2-acetyl precursor and the loss of C2H4 from the 2-propyl precursor are virtually identical. A representative spectrum is shown in Fig. 4(a). It is clearly characteristic of the 2-methylene structure 9 䡠⫹ and,

see Fig. 4(c), considerably different from that of the 2-methyl-pyrazine isomer 8 䡠⫹. For 9 䡠⫹ the structure diagnostic signals at m/z 80 and 47 (charge stripping peak) as well as the characteristic m/z 26:28 peak intensity ratio are reproduced in the CID spectrum, Fig. 4(b), of the intense survivor ion in the NR spectrum. Not only does this attest to the isomeric purity of the decarbonylation product, it also shows that 2-methylene-1,2-dihydropyrazine, 9, is a stable, noninterconverting species in the dilute gas phase. For ion 8 䡠⫹ the NR/CID and CID spectra are also closely similar and thus the postcollisional isomerization effect observed for 2-methylpyridine is clearly absent in the pyrazine analogue. Since the m/z 67 peak in the CID spectra contains a large metastable ion (MI) contribution, it is not considered in the comparisons. Another pyrazine ylide ion, 10 䡠⫹, was also probed for the stability of its neutral counterpart. The ion was generated from the dissociative ionization of methyl5-methylpyrazine-2-carboxylate, VII 䡠⫹, as depicted in Scheme 4 [36]. A distinct CID mass spectrum was obtained, see Fig. 4(e), which is closely similar, see Fig. 4(f), to the CID spectrum of the intense survivor

D.J. Lavorato et al./International Journal of Mass Spectrometry 195/196 (2000) 71– 83

81

Fig. 4. NR/CID and comparative CID mass spectra (3ffr, 10 keV ions, collision gas O2) of C5H6N2䡠⫹ (m/z 94) ions. (a), (b) CID and NR/CID spectra of C5H6N2䡠⫹ from ionized 2-acetylpyrazine or 2-propylpyrazine; (c), (d) CID and NR/CID spectra of ionized methylpyrazine; (e), (f): CID and NR/CID spectra of C5H6N2䡠⫹ from ionized methyl-5-methylpyrazine-2-carboxylate. Spectra (e) and (f) are dominated by a peak at m/z 93, which is not shown because it is largely of metastable origin.

peak in the NR spectrum. This leaves little doubt that 5-methyl-1,2-dihydropyrazine, 10, is a stable species in the gas phase.

For the thiazole system, the decarbonylation of the 2-acetyl precursor and the C2H4 elimination from the 2-propyl precursor again yielded product ions of the

Scheme 4.

82

D.J. Lavorato et al./International Journal of Mass Spectrometry 195/196 (2000) 71– 83

Fig. 5. CID mass spectra (10 keV ions, collision gas O2) of C4H5NS 䡠⫹ (m/z 99) ions generated from ionized (a) 2-acetylthiazole or 2-propylthiazole and (b) 2-methylthiazole.

same structure, as witnessed by their superimposable CID mass spectra. A representative spectrum is shown in Fig. 5(a). Unlike the situation with the pyridine and pyrazine analogues, this spectrum is markedly different from that of the 2-methyl substituted isomer 11 䡠⫹, see Fig. 5(b). The CID mass spectrum of the decarbonylation product ion is characterized by unique peaks at m/z 71 (loss of CH2N 䡠), m/z 59 (loss of C2HN), m/z 54 (loss of CHS 䡠), and an intense charge stripping peak at m/z 49.5. These CID features are compatible with the expected 2-methylene-1,2-dihydrothiazole structure 12 䡠⫹ but they do not necessarily rule out that the structurally related 2-methylene isomer 13 䡠⫹ is (co)-generated in the decarbonylation. However, the computational results indicate that this is unlikely. Not only is 13 䡠⫹ found to undergo ring opening upon geometry optimization, but the resulting minimum energy structure, CH2¢C¢N–C(H)¢C(H)SH 䡠⫹ lies 35 kcal/mol higher in energy than 12 䡠⫹. The N-methyl and S-methyl ylide isomers 14 䡠⫹ and 15 䡠⫹ retain their cyclic structure upon geometry optimization, but these species are even higher in energy, see Table 1. Thus, we conclude that the decarbonylation of 2-acetylthiazole occurs as established for the pyridine and pyrazine analogues and yields 12 䡠⫹ as the product ion. A NR experiment yielded intense “survivor” ions. Their CID spectrum appeared to be indistinguishable from that of the source generated ions (spectra not shown), confirming that 2-methylene-2,3-dihydrothiazole, 12, is also a stable species in the rarefied gas phase.

Acknowledgements The work at McMaster University was supported by the Natural Sciences and Engineering Research Council of Canada (NSERC). One of the authors (D.J.L.) would like to thank NSERC for a Post Graduate Scholarship. Another author (G.A.M.) would also like to thank NSERC for a Post Doctoral Fellowship and the Deutscher Akademischer Austauschdienst for a Research Scholarship. The research at the TU Berlin was funded by the Deutsche Forschungsgemeinschaft and the Fonds der Chemischen Industrie.

References [1] K.L. Busch, G.L. Glish, S.A. McLuckey, Mass Spectrometry/ Mass Spectrometry, Techniques and Applications of Tandem Mass Spectrometry, VCH, New York, 1988. [2] (a) K. Biemann, Mass Spectrometry: Organic Chemical Applications, McGraw-Hill, New York, 1962; (b) G. Spiteller, Advances in Heterocyclic Chemistry, Academic, New York, 1966, Vol. 7; (c) K.R. Jennings, J.H. Futrell, J. Chem. Phys. 44 (1966) 4115; (d) W.G. Cole, D.H. Williams, A.N.H. Yeo, J. Chem. Soc. B (1968) 1284; (e) N.M.M. Nibbering, T.J. de Boer, Org. Mass Spectrom. 3 (1970) 597. [3] For review see A. Padwa, S.F. Hornbuckle, Chem. Rev. 91 (1991) 298. [4] C.G. De Koster, J.J. van Houte, J. van Thuijl, Int. J. Mass Spectrom. Ion Processes 98 (1990) 235. [5] R. Flammang, O. Thoelen, C. Quattrocchi, J.-L. Bredas, Rapid Commun. Mass Spectrom. 6 (1992) 135. [6] S. Yu, M. Gross, K. Fountain, J. Am. Soc. Mass Spectrom. 4 (1993) 117. [7] For a retrospective see R.G. Cooks, J. Mass Spectrom. 30 (1995) 1215.

D.J. Lavorato et al./International Journal of Mass Spectrometry 195/196 (2000) 71– 83 [8] For some recent reviews see (a) F.W. McLafferty, Int. J. Mass Spectrom. Ion Processes 118/119 (1992) 211; (b) F. Turecˆek, Org. Mass Spectrom. 27 (1992) 1087; (c) N. Goldberg, H. Schwarz, Acc. Chem. Res. 27 (1994) 347; (d) C.A. Schalley, G. Hornung, D. Schro¨der, H. Schwarz, Chem. Soc. Rev. 27 (1998) 91; (e) C.A. Schalley, G. Hornung, D. Schro¨der, H. Schwarz, Int. J. Mass Spectrom. Ion Processes 172 (1998) 181. [9] (a) P.D. Michie, B.M. Johnson, Org. Mass Spectrom. 18 (1983) 202; (b) C.P. Whittle, M.J. Lacey, C.G. MacDonald, ibid. 11 (1976) 848; (c) D.L. von Minden, J.G. Liehr, M.H. Wilson, J.A. McCloskey, J. Org. Chem. 39 (1974) 285; (d) N. Neuner-Jehle, Tetrahedron Lett. (1968) 2047; (e) C.P. Whittle, Tetrahedron Lett. (1968) 3689; (f) Y. Rahamin, J. Sharvit, A. Mandelbaum, P. Sprecher, J. Org. Chem. 32 (1967) 3856. [10] H. Budzikiewicz, E. Lange, W. Ockels, Phosphorus Sulfur 11 (1981) 33. [11] (a) A. Ferretti, V.P. Flanagan, J. Agr. Food Chem. 19 (1971) 245; (b) M.D. Migahed, F.A. Abd El-Kader, Int. J. Mass Spectrom. Ion Phys. 31 (1979) 373. [12] K.B. Tomer, C. Djerassi, J. Org. Chem. 38 (1973) 4152. [13] (a) R.J. Moser, E.V. Brown, Org. Mass Spectrom. 4 (1970) 555; (b) E.V. Brown, R.J. Moser, J. Heterocycl. Chem. 8 (1971) 189. [14] P.H. Chen, J. Org. Chem. 419 (1976) 2973. [15] D.J. Lavorato, J.K. Terlouw, T.K. Dargel, W. Koch, G.A. McGibbon, H. Schwarz, J. Am. Chem. Soc. 118 (1996) 11898, and references cited therein. [16] (a) J.J. Brophy, G.W.K. Cavill, Heterocycles 14 (1990) 477; (b) J.A. Maga, C.E. Sizer, Fernarolis Handbook of Flavour Ingredients, CRC, Cleveland, OH, 1975, Vol. 1. [17] G.M. Clarke, R. Grigg, D.H. Williams, J. Chem. Soc. B (1966) 339. [18] H.F. van Garderen, P.J.A. Ruttink, P.C. Burgers, G.A. McGibbon, J.K. Terlouw, Int. J. Mass Spectrom. Ion Processes 121 (1992) 159. [19] F. Krohnke, Chem. Ber. 70 (1937) 543. [20] J.D. Behun, R. Levine, J. Org. Chem. 26 (1961) 3381. [21] D. Pitre`, S. Boveri, E.B. Grabitz, Chem. Ber. 99 (1966) 364. [22] GAUSSIAN 94 (Revision B.2): M.J. Frisch, G.W. Trucks, H.B. Schlegel, P.M.W. Gill, B.G. Johnson, M.A. Robb, J.R. Cheeseman, T.A. Keith, G.A. Peterson, J.A. Montgomery, K. Raghavachari, J.B. Foresman, J. Cioslowski, B.B. Stefanov,

[23] [24] [25] [26]

[27]

[28] [29]

[30] [31] [32] [33]

[34] [35] [36]

83

A. Nanayakkara, M. Challacombe, C.Y. Peng, P.Y. Ayala, W. Chen, M.W. Wong, J.L. Andres, E.S. Replogle, R. Gomperts, R.L. Martin, D.J. Fox, J.S. Binkley, D.J. DeFrees, J. Baker, J.J.P. Stewart, M. Head-Gordon, C. Gonzalez, J.A. Pople, Gaussian, Inc., Pittsburgh PA, 1995. (a) A.D. Becke, J. Phys. Chem. 98 (1993) 5648; (b) C. Lee, W. Yang, R.G. Parr, Phys. Rev. B 37 (1988) 785. (a) A.P. Scott, L. Radom, J. Phys. Chem. 100 (1996) 16502; (b) M.W. Wong, Chem. Phys. Lett. 256 (1996) 391. D.V. Bowen, P.W. Skett, J. Thorpe, A.O. Plunkett, Org. Mass Spectrom. 19 (1984) 285. The CID spectra reported in the present study are not identical to those reported in [5]. However, a discussion of the variations that may be attributed to differing instrumental and experimental conditions would form an unnecessary digression in the context of this work. D.J. Lavorato, J.K. Terlouw, G.A. McGibbon, T.K. Dargel, W. Koch, H. Schwarz, Int. J. Mass Spectrom. Ion Processes 179/180 (1998) 7. M. Tiecco, Acc. Chem. Res. 18 (1981) 16. ¥⌬H f [pyridine-2-ylidene] 䡠⫹ ⫹ ⌬H f [CH2C¢O] ⫽ 237 ⫺ 11.4 ⫽ 225.6 kcal/mol [15,30] whereas ¥⌬H f [6 䡠⫹] ⫹ ⌬H f [CO] ⫽ 208 ⫺ 26.4 ⫽ 181.6 kcal/mol [30]. ⌬H f [6 䡠⫹] was estimated to be 208 kcal/mol, based on ⌬H f [2 䡠⫹] ⫽ 232 kcal/mol [30] and the difference in the relative energies of 6 䡠⫹ and 2 䡠⫹ presented in Table 1. S. Lias, J.E. Bartmess, J.F. Liebman, J.L. Holmes, R.D. Levin, W.G. Mallard, J. Phys. Chem. Ref. 17, 1988, Suppl. 1. W. Heerma, M.M. Sarneel, G. Dijkstra, Org. Mass Spectrom. 21 (1986) 681. R. Neeter, N.M.M. Nibbering, Tetrahedron 28 (1972) 2575. The MI spectrum of 2 䡠⫹ shows that the reactions of lowest energy requirement are loss of HCN (m/z 66), C2H2 (m/z 67), and H 䡠 (m/z 92). The activation energy for the latter process has been established as 77 kcal/mol, from an appearance energy measurement [34]. T.F. Palmer, F.P.L. Lossing, J. Am. Chem. Soc. 85 (1963) 1733. S. Beranova, C. Wesdemiotis, J. Am. Soc. Mass Spectrom. 5 (1994) 1093. T.K. Dargel, W. Koch, D.J. Lavorato, G.A. McGibbon, J.K. Terlouw, H. Schwarz, Int. J. Mass Spectrom. Ion Processes 185/186/187 (1999) 925.